As i continue to get used to using my energy monitoring setup i now have power & kWh from my three CT’s (single phase, 2 on Line before & after the meter one on Neutral) and I’m tracking meter pulsecounts which i’m scaling (1000 per kWh) to give me the meter kWh usage.
Over the few days that i’ve had this running, my 3 CT’s are tracking pretty closely (within 10W), but i’m getting an ever growing divergence between the pulsecount and the CT’s - currently in the order of approx 0.5kWh for 17.5kWh usage over 24 hours.
I get the need for calibration over nominal calculated coefficients to account for manufacturing tolerances and individual setups and will be calibrating my system in due course. I am curious before i do though as to what to expect - i.e how close should i be able to get my CT / Vrms calculated kWh to the pulsecount value over the long term? Is it possible to eradicate this divergence or will there always be some drift?
And even then, given that you are unlikely to have a multimeter that’s better than 1% accuracy, and because c.t. and v.t (a.c. adapter) errors both depend on the magnitude of the quantity, you’ll still need to tweak the calibration over time to get the best tracking against the pulse count. And because of that, expect it to deviate according to the loads you have on at any particular time.
Even a smallish positive offset on your CT will lead to a meaningful cumulative drift e.g. if you consistently read 10W over the real value that will give 0.24kWh per day.
You can only tell that by examining the Feeds. For me the CT reads more than the Pulse.
No because you were seeing a figure where it should be zero.
Yes - although it does depend on your definition of meaningful and what error you are willing to accept. Remember it is a % difference not a fixed offset and that % is not linear as the measurements are less accurate at low power. (if I’m wrong @Robert.Wall will correct that).
By calibrating the CT against the pulse, I have it to less than 0.5% measured over 290Hrs. This will fluctuate slightly with Voltage Frequency (AIUI).
You can add the load current and its power factor to that list.
That will depend entirely on the particular set of components you have in your emonPi/emonTx that affect the particular channel you’re comparing. The calibration constants have been set assuming the nominal values for all components, whether any particular input reads high or low is a lottery. But - if all the tolerances that are common to both/all four channels go the same way, then all will tend to read wrongly in the same direction.
Here’s my present test with an emonPi - not calibrated particularly carefully, after 38 kWh:
The ‘blue’ c.t. is 0.66% low, the ‘red’ c.t. is about 3.5% low over the period, but looking at the powers,
they are 1 or 2 W different at 5.5 kW. So that’s the c.t. (mostly phase) errors at low power. The ‘blue’ c.t. is a recent moulded plug one, the ‘red’ one is an old grey-lead one.
